Version 1
: Received: 30 April 2017 / Approved: 1 May 2017 / Online: 1 May 2017 (10:44:08 CEST)
Version 2
: Received: 15 November 2017 / Approved: 16 November 2017 / Online: 16 November 2017 (04:29:19 CET)
Li, H.; Zheng, H.; Han, C.; Wang, H.; Miao, M. Onboard Spectral and Spatial Cloud Detection for Hyperspectral Remote Sensing Images. Remote Sensing 2018, 10, 152, doi:10.3390/rs10010152.
Li, H.; Zheng, H.; Han, C.; Wang, H.; Miao, M. Onboard Spectral and Spatial Cloud Detection for Hyperspectral Remote Sensing Images. Remote Sensing 2018, 10, 152, doi:10.3390/rs10010152.
Li, H.; Zheng, H.; Han, C.; Wang, H.; Miao, M. Onboard Spectral and Spatial Cloud Detection for Hyperspectral Remote Sensing Images. Remote Sensing 2018, 10, 152, doi:10.3390/rs10010152.
Li, H.; Zheng, H.; Han, C.; Wang, H.; Miao, M. Onboard Spectral and Spatial Cloud Detection for Hyperspectral Remote Sensing Images. Remote Sensing 2018, 10, 152, doi:10.3390/rs10010152.
Abstract
It is strongly desirable to accurately detect the clouds in hyperspectral images onboard before compression. However, conventional onboard cloud detection methods are not appropriate to all situation such as shadowed cloud or darken snow covered surfaces which are not identified properly in the NDSI test. In this paper, we propose a new spectral–spatial classification strategy to enhance the orbiting cloud screen performances obtained on hyperspectral images by integrating threshold assisted exponential spectral angle map (TESAM), adaptive Markov random fields (aMRFs) and dynamic stochastic resonance (DSR). First, TESAM is performed to classify the cloud pixels coarsely based on spectral information. Then aMRFs is performed to do optimal process by using spatial information, which improved the classification performance significantly. But some misclassification points still exist after aMRFs processing because of the noise of data in the onboard environment. DSR is used to eliminate misclassification points in binary labeling image after aMRFs. Taking level 0.5 data from hyperion as dataset, the average overall accuracy of the proposed algorithm is 96.28% after test. The method can provide an accurate cloud mask for the on-going EO-1 images and the similar satellites with the same spectral settings without manual intervention. The experiment indicate that the proposed method reveals better performance than the classical onboard cloud detection or current advanced hyperspectral classification methods.
Keywords
onboard cloud detecion; region of interesting compression; themodynamic phase; spectral angle map; markov random field; dynamic stochastic resonance
Subject
Environmental and Earth Sciences, Atmospheric Science and Meteorology
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.